load("SHOC-humain.RData")
# size
# sort(sapply(ls(), function(x) object.size(get(x))), decreasing = TRUE)
# alphabet
# obj_sizes <- sapply(ls(), function(x) object.size(get(x)))
# sorted_names <- names(sort(obj_sizes)) # Tri par nom (ordre alphabétique)
# obj_sizes[sorted_names]
# rm(m_full_2.1,m_full_2,m_full_4,m_full_3,m_full_2.2,m_full_1)
# rm(list = ls(pattern = "^niv1_fixed_"))
# save.image(file = "SHOC-humain.RData")SHOC-humain
DATA
quiz
participant.es
Sexe
Femme, homme, NA, en fonction des prénoms détectés dans les emails.
Cocheurs
Correspondance des emails entre les répondant.es au quiz et les 1741 cocheurs sur cocheurs.fr
Ancienneté
Ancienneté déclaré par les répondant.es pour le.s suivis participatifs
Departement
Département français métropolitain + suisse
01 02 03 04 05 06 07 08 09 10 11 13 14 15 16 17
10 2 2 2 4 2 3 2 3 6 4 4 4 1 2 7
18 19 21 22 23 24 25 26 27 28 29 2B 30 31 32 33
10 1 6 5 4 8 2 6 2 4 8 2 7 12 5 15
34 35 37 38 40 41 42 43 44 45 46 47 48 49 50 51
10 7 7 7 2 6 13 2 23 7 7 2 1 16 2 3
52 53 55 56 57 58 59 60 62 63 64 65 66 67 68 69
3 7 1 9 1 1 11 5 10 3 3 1 7 4 7 10
70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85
1 7 5 2 3 4 4 3 13 17 5 6 1 6 6 10
86 87 88 89 90 91 92 93 94 95 su <NA>
4 2 1 3 2 7 10 5 3 5 2 1
Région et biorégion
Départements associés à une région admin, à une biorégion (INPN)
Bagueurs
Correspondances des emails entre les participant.es et les 32 premiers bagueurs (CRBPO)
Expert sciences participatives
Si participation_stoc == “Oui”, participation_shoc == “Oui”, participation_epoc == “Oui”, bagueur == “Oui” ou nb_obs_cocheurs >= 300, alors la·e participant·e est considéré·e comme expert·e.
Species
Nom vernatuclaire et code espèce
Shoc
Tendances SHOC
Augmentation forte Augmentation modérée
16 4
Augmentation modérée à forte Déclin fort
9 4
Déclin modéré Déclin modéré à fort
8 2
Incertain Stable
53 12
Stoc
Tendances SHOC et STOC
traits
catégories de tendances
[1] "avis quiz shoc"
En augmentation En déclin Fluctuante Je ne sais pas Stable
2146 3496 1139 2501 5208
<NA>
0
Fort Modéré Modéré à fort <NA>
681 3095 1519 9195
Augmentation Déclin Fluctuante Incertain Stable <NA>
2146 3496 1139 2501 5208 0
Augmentation Augmentation fort
194 179
Augmentation modéré Augmentation modéré à fort
1323 450
Déclin Déclin fort
192 501
Déclin modéré Déclin modéré à fort
1744 1059
Fluctuante Fluctuante modéré
1132 7
Incertain Incertain modéré
2499 1
Incertain modéré à fort Stable
1 5178
Stable fort Stable modéré
1 20
Stable modéré à fort <NA>
9 0
Augmentation Augmentation fort
194 179
Augmentation modéré Augmentation modéré à fort
1323 450
Déclin Déclin fort
192 501
Déclin modéré Déclin modéré à fort
1744 1059
Fluctuante Incertain
1139 2501
Stable <NA>
5208 0
[1] "catégoies quiz shoc après transformation en 2 nouveaux"
[1] "niveau 1"
Augmentation Déclin Fluctuante Incertain Stable
2146 3496 1139 2501 5208
[1] "niveau 2"
Augmentation fort Augmentation modéré
179 1323
Augmentation modéré à fort Déclin fort
450 501
Déclin modéré Déclin modéré à fort
1744 1059
Fluctuante Incertain
1139 2501
Stable
5208
[1] "catégoies des tendances shoc"
[1] "niveau 1"
Augmentation Déclin Incertain Stable
29 14 53 12
[1] "niveau 2"
Augmentation fort Augmentation modéré
16 4
Augmentation modéré à fort Déclin fort
9 4
Déclin modéré Déclin modéré à fort
8 2
Incertain Stable
53 12
dataset
comparaison SHOC vs Humain
Si tendances SHOC et réponses identiques ou différentes
Niveau 1 :
Augmentation = augmentation
Déclin = déclin
Stable = stable
Incertain = je ne sais pas, fluctuante ou incertain
[1] "Jeu de données propres à analyser"
exploration données
email espece avis_niv1 avis_niv2
Length:13720 Length:13720 Length:13720 Length:13720
Class :character Class :character Class :character Class :character
Mode :character Mode :character Mode :character Mode :character
migrateur migrateur_simplifie regime abondance
Length:13720 Length:13720 Length:13720 Min. : 1500
Class :character Class :character Class :character 1st Qu.: 281250
Mode :character Mode :character Mode :character Median : 775000
Mean :1685679
3rd Qu.:1862500
Max. :9000000
detectabilite gregaire fluctuante popularite
Length:13720 Length:13720 Length:13720 Length:13720
Class :character Class :character Class :character Class :character
Mode :character Mode :character Mode :character Mode :character
fauneFrance abondance_log departement participation_stoc
Min. :-0.852880 Min. : 7.313 Length:13720 Length:13720
1st Qu.:-0.636089 1st Qu.:12.540 Class :character Class :character
Median :-0.452104 Median :13.560 Mode :character Mode :character
Mean : 0.007713 Mean :13.334
3rd Qu.: 0.093103 3rd Qu.:14.420
Max. : 2.435363 Max. :16.013
participation_shoc participation_epoc participation_odj sexe
Length:13720 Length:13720 Length:13720 Length:13720
Class :character Class :character Class :character Class :character
Mode :character Mode :character Mode :character Mode :character
cocheur nb_obs_cocheurs anciennete region
Length:13720 Min. :111.0 Min. : 0.000 Length:13720
Class :character 1st Qu.:265.0 1st Qu.: 0.000 Class :character
Mode :character Median :344.5 Median : 4.000 Mode :character
Mean :334.3 Mean : 5.857
3rd Qu.:415.0 3rd Qu.:10.000
Max. :482.0 Max. :41.000
NA's :12264
bioregion bagueur expert tendance_shoc
Length:13720 Length:13720 Length:13720 Length:13720
Class :character Class :character Class :character Class :character
Mode :character Mode :character Mode :character Mode :character
tendances_stoc tendances_stoc_simplifie pourcentage_stoc
Length:13720 Length:13720 Min. :-0.77334
Class :character Class :character 1st Qu.:-0.43732
Mode :character Mode :character Median :-0.24471
Mean :-0.09049
3rd Qu.:-0.08037
Max. : 2.77736
NA's :490
tendance_shoc_niv1 tendance_shoc_niv2 comparaison_niv1 comparaison_niv2
Length:13720 Length:13720 Length:13720 Length:13720
Class :character Class :character Class :character Class :character
Mode :character Mode :character Mode :character Mode :character
[1] "espece"
Accenteur mouchet Alouette des champs Bouscarle de Cetti
490 490 490
Bruant zizi Chardonneret élégant Choucas des tours
490 490 490
Geai des chênes Grimpereau des jardins Grive draine
490 490 490
Grive litorne Grosbec cassenoyaux Merle noir
490 490 490
Mésange bleue Mésange noire Mésange nonnette
490 490 490
Pic vert Pie bavarde Pigeon ramier
490 490 490
Pinson des arbres Pipit farlouse Pouillot véloce
490 490 490
Roitelet à triple bandeau Roitelet huppé Rougegorge familier
490 490 490
Tarin des aulnes Tourterelle turque Troglodyte mignon
490 490 490
Verdier d'Europe <NA>
490 0
[1] "departement"
01 02 03 04 05 06 07 08 09 10 11 13 14 15 16 17
280 56 56 56 112 56 84 56 84 168 112 112 112 28 56 196
18 19 21 22 23 24 25 26 27 28 29 2B 30 31 32 33
280 28 168 140 112 224 56 168 56 112 224 56 196 336 140 420
34 35 37 38 40 41 42 43 44 45 46 47 48 49 50 51
280 196 196 196 56 168 364 56 644 196 196 56 28 448 56 84
52 53 55 56 57 58 59 60 62 63 64 65 66 67 68 69
84 196 28 252 28 28 308 140 280 84 84 28 196 112 196 280
70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85
28 196 140 56 84 112 112 84 364 476 140 168 28 168 168 280
86 87 88 89 90 91 92 93 94 95 su <NA>
112 56 28 84 56 196 280 140 84 140 56 0
[1] "participation_stoc"
Non Oui <NA>
7840 5880 0
[1] "participation_shoc"
Non Oui <NA>
8176 5544 0
[1] "participation_epoc"
Non Oui <NA>
9492 4228 0
[1] "sexe"
F H <NA>
2856 8512 2352
[1] "cocheur"
Non Oui <NA>
12264 1456 0
[1] "anciennete"
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
3640 644 980 1148 924 1120 504 392 504 140 1400 280 336 84 28 448
17 18 19 20 21 22 23 24 25 41 <NA>
84 140 28 392 28 56 168 112 112 28 0
[1] "region"
Auvergne-Rhône-Alpes Bourgogne-Franche-Comté
2016 1092
Bretagne Centre-Val de Loire
644 1288
Corse Grand Est
56 952
Hauts-de-France Île-de-France
728 1176
Normandie Nouvelle-Aquitaine
1120 1792
Occitanie Pays de la Loire
1820 448
Provence-Alpes-Côte d’Azur su
532 56
<NA>
0
[1] "bioregion"
ALP ATL CON MATL MED MMED <NA>
1092 2072 6776 2408 476 896 0
[1] "bagueur"
Non Oui <NA>
12796 924 0
[1] "expert"
Non Oui <NA>
5124 8596 0
[1] "tendances_stoc"
Augmentation forte Augmentation modérée
1470 2940
Augmentation modérée à forte Déclin fort
490 980
Déclin modéré Déclin modéré à fort
3430 1470
Stable <NA>
2450 490
[1] "pourcentage_stoc"
-0.773339424601803 -0.664473778519284 -0.662380208402312 -0.649818787700483
490 490 490 490
-0.571309908314051 -0.523157795623706 -0.437321420827873 -0.41847928977513
490 490 490 490
-0.325315419569897 -0.320081494277468 -0.286584372405923 -0.28239723217198
490 490 490 490
-0.261461531002265 -0.244712970066493 -0.210169063136463 -0.177718726323404
490 490 490 490
-0.155736240095203 -0.132706968808516 -0.125379473399116 -0.0992098469369719
490 490 490 490
-0.0803677158842282 -0.0165138273165966 0.0389657807831488 0.12270858546201
490 490 490 490
0.600042572131518 1.43642383386164 2.77735549378191 <NA>
490 490 490 490
[1] "tendances_stoc_simplifie"
Augmentation Déclin Stable <NA>
4900 5880 2450 490
[1] "tendances_stoc_simplifie_2"
Warning: Unknown or uninitialised column: `tendances_stoc_simplifie_2`.
<NA>
0
[1] "detectabilite"
Faible Forte <NA>
6860 6860 0
[1] "gregaire"
Non Oui <NA>
5880 7840 0
[1] "fluctuante"
Non Oui <NA>
10290 3430 0
[1] "migrateur"
Non Oui Partiel <NA>
4410 1470 7840 0
[1] "migrateur simplifié"
Non Oui <NA>
4410 9310 0
[1] "abondance"
1500 22500 45000 75000 80000 225000 3e+05 4e+05 525000 650000
490 490 490 490 490 980 490 490 490 490
7e+05 750000 8e+05 1250000 1300000 1500000 1650000 2500000 3250000 4e+06
980 490 490 490 490 1470 490 490 980 490
4500000 6500000 9e+06 <NA>
490 490 490 0
[1] "abondance log"
7.3132203870903 10.0212705881925 10.7144177687525 11.2252433925184
490 490 490 490
11.289781913656 12.3238556811866 12.6115377536383 12.8992198260901
490 980 490 490
13.1711535415738 13.3847276418718 13.4588356140255 13.5278284855125
490 490 980 490
13.5923670066501 14.0386541092785 14.0778748224318 14.2209756660724
490 490 490 1470
14.3162858458768 14.7318012898384 14.9941655543059 15.2018049190842
490 490 980 490
15.3195879547405 15.6873127348659 16.0127351353005 <NA>
490 490 490 0
[1] "fauneFrance"
-0.852879689044691 -0.840494517413959 -0.838284301897863 -0.811244007565813
490 490 490 490
-0.787270138364202 -0.718195925523346 -0.717419363314988 -0.608979420065839
490 490 490 490
-0.570290692608421 -0.561429918692543 -0.544026960484729 -0.520869477194468
490 490 490 490
-0.494127860634863 -0.476247017991137 -0.427960778112476 -0.376926702727318
490 490 490 490
-0.369260639901221 -0.281389639247807 -0.0582773432311487 -0.0549719758827538
490 490 490 490
0.0410429720326701 0.249281114981551 1.13065930961621 1.16377271865465
490 490 490 490
1.85897510179839 2.00526747474211 2.24215877199424 2.43536346706337
490 490 490 490
<NA>
0
[1] "regime"
Granivore Insectivore Omnivore <NA>
4410 4410 4900 0
[1] "popularite"
Non Oui <NA>
5390 8330 0
[1] "comparaison_niv1"
different identique <NA>
9436 4284 0
[1] "comparaison_niv2"
different identique <NA>
10500 2864 356
FREQUENCES
niv 1
Shoc vs hasard
Chi-squared test for given probabilities
data: obs_shoc
X-squared = 700, df = 3, p-value < 2.2e-16
Tendance Observé Attendu Résidu_std.Var1 Résidu_std.Freq
Augmentation Augmentation 4410 3430 Augmentation 19.321836
Déclin Déclin 3920 3430 Déclin 9.660918
Incertain Incertain 2940 3430 Incertain -9.660918
Stable Stable 2450 3430 Stable -19.321836
p-value < 0.05 → la distribution est significativement différente du hasard
Niv 1 vs hasard
Chi-squared test for given probabilities
data: obs_avis_niv1
X-squared = 3238.7, df = 4, p-value < 2.2e-16
Tendance Observé Attendu Résidu_std.Var1 Résidu_std.Freq
Augmentation Augmentation 2033 2744 Augmentation -15.17513
Déclin Déclin 3269 2744 Déclin 11.20527
Fluctuante Fluctuante 1054 2744 Fluctuante -36.07029
Incertain Incertain 2361 2744 Incertain -8.17451
Stable Stable 5003 2744 Stable 48.21467
p-value < 0.05 → la distribution est significativement différente du hasard
Shoc vs niv 1
Augmentation Déclin Incertain Stable
SHOC 4410 3920 2940 2450
AVIS 2033 3269 2361 5003
Pearson's Chi-squared test
data: tab_chi2
X-squared = 1834.5, df = 3, p-value < 2.2e-16
[1] "résidus"
Augmentation Déclin Incertain Stable
SHOC 30.39867 5.03482 5.647143 -39.01322
AVIS -30.39867 -5.03482 -5.647143 39.01322
[1] "proportions"
Augmentation Déclin Incertain Stable
SHOC 0.3214286 0.2857143 0.2142857 0.1785714
AVIS 0.1605084 0.2580925 0.1864045 0.3949945
p-value < 0.05 → les distributions diffèrent significativement entre SHOC et AVIS
Conclusions :
🔹 Augmentation
Résidus : SHOC +27.6, AVIS −27.6
Proportions : SHOC 30.0 %, AVIS 16.1 %
👉 Augmentation est massivement plus fréquente dans SHOC que dans AVIS C’est l’un des moteurs principaux du χ².
🔹 Déclin
Résidus : SHOC +0.92, AVIS −0.92
Proportions : SHOC 26.7 %, AVIS 26.2 %
👉 Aucune différence réelle Déclin se comporte de la même manière dans les deux sources.
🔹 Incertain
Résidus : SHOC +16.1, AVIS −16.1
Proportions : SHOC 26.7 %, AVIS 18.6 %
👉 Incertain est fortement sur-représenté dans SHOC Clairement différent entre les deux distributions.
🔹 Stable
Résidus : SHOC −42.2, AVIS +42.2
Proportions : SHOC 16.7 %, AVIS 39.1 %
👉 Stable est massivement plus fréquent dans AVIS que dans SHOC C’est le contraste le plus fort du tableau.
En résumé :
Les distributions de tendances diffèrent fortement entre SHOC et AVIS (χ², p < 0.001). Les écarts sont principalement dus à une sur-représentation des modalités « Augmentation » et « Incertain » dans SHOC, tandis que la modalité « Stable » est largement dominante dans AVIS. La modalité « Déclin » ne montre pas de différence significative entre les deux sources.
SHOC → profil plus dynamique / incertain : plus d’Augmentation, plus d’Incertain, peu de Stable
AVIS → profil nettement plus stable, Stable ≈ 40 %, moins d’Augmentation
👉 Ce n’est pas un effet d’échantillon, mais un changement structurel de profil.
Interprétations possibles ?
Chez les participant.es, biais vers le stable, et peu d’aumgmentation. En demandant de répondre pour des tendances, on pense moins facilement à répondre stable ? Et oin parle beaucoup des déclins (média, crise ecolo…), mais peu des augmentations, donc les gens n’y pensent pas ?
!!! niv 2
PROPORTION IDENTIQUE vs DIFFERENT
niv 1
[1] "proportion de tendances d'identiques vs différentes, niveau 1"
[1] "toutes les tendances"
Var1 Freq prop
1 different 9436 0.6877551
2 identique 4284 0.3122449
3 <NA> 0 0.0000000
[1] "declin et augmentation seulement"
Var1 Freq prop
1 different 5964 0.7159664
2 identique 2366 0.2840336
3 <NA> 0 0.0000000
!!! niv 2
[1] "proportion de tendances d'identiques vs différentes, niveau 2"
[1] "toutes les tendances"
Var1 Freq prop
1 different 10500 0.76530612
2 identique 2864 0.20874636
3 <NA> 356 0.02594752
[1] "declin et augmentation seulement"
Var1 Freq prop
1 different 5964 0.7159664
2 identique 2366 0.2840336
3 <NA> 0 0.0000000
ANALYSES ORDINALES
niv 1
données
aléatoire
df AIC
m_n1.1 3 14300.13
m_n1.7 4 14302.11
m_n1.5 4 14302.13
m_n1.6 4 14302.13
m_n1.4 3 14442.81
m_n1.2 3 14468.69
m_n1.3 3 14473.24
Cumulative Link Mixed Model fitted with the Laplace approximation
formula: avis_niv1 ~ 1 + (1 | email)
data: dt1
link threshold nobs logLik AIC niter max.grad cond.H
logit flexible 7120 -7147.06 14300.13 117(368) 3.27e-06 7.3e+00
Random effects:
Groups Name Variance Std.Dev.
email (Intercept) 0.2729 0.5224
Number of groups: email 397
No Coefficients
Threshold coefficients:
Estimate Std. Error z value
Déclin|Stable -1.17264 0.03910 -29.99
Stable|Augmentation 1.33043 0.03986 33.37
base
Cumulative Link Mixed Model fitted with the Laplace approximation
formula: avis_niv1 ~ tendance_shoc_niv1 + (1 | email)
data: dt1
link threshold nobs logLik AIC niter max.grad cond.H
logit flexible 7120 -6727.49 13464.97 222(682) 3.84e-03 6.3e+00
Random effects:
Groups Name Variance Std.Dev.
email (Intercept) 0.3294 0.5739
Number of groups: email 397
Coefficients:
Estimate Std. Error z value Pr(>|z|)
tendance_shoc_niv1.L 1.1356 0.0408 27.834 < 2e-16 ***
tendance_shoc_niv1.Q -0.1313 0.0442 -2.971 0.00297 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Threshold coefficients:
Estimate Std. Error z value
Déclin|Stable -1.27787 0.04243 -30.11
Stable|Augmentation 1.48848 0.04349 34.23
espece
Cumulative Link Mixed Model fitted with the Laplace approximation
formula: avis_niv1 ~ tendance_shoc_niv1 * espece + (1 | email)
data: dt1
link threshold nobs logLik AIC niter max.grad cond.H
logit flexible 7120 -5996.93 12041.85 3714(11170) 3.76e-03 1.4e+02
Random effects:
Groups Name Variance Std.Dev.
email (Intercept) 0.4857 0.6969
Number of groups: email 397
Coefficients:
Estimate Std. Error z value Pr(>|z|)
tendance_shoc_niv1.L 1.01063 0.10757 9.395 < 2e-16 ***
tendance_shoc_niv1.Q -0.94924 0.10793 -8.795 < 2e-16 ***
especeBouscarle de Cetti 1.02421 0.16866 6.072 1.26e-09 ***
especeBruant zizi -0.98578 0.16540 -5.960 2.53e-09 ***
especeChoucas des tours 1.07236 0.15576 6.885 5.79e-12 ***
especeGeai des chênes 1.46891 0.15551 9.446 < 2e-16 ***
especeGrimpereau des jardins -0.64779 0.15154 -4.275 1.91e-05 ***
especeGrive draine -0.49593 0.16461 -3.013 0.00259 **
especeMerle noir 1.13181 0.15489 7.307 2.73e-13 ***
especeMésange bleue -0.01333 0.14946 -0.089 0.92895
especeMésange nonnette -2.35238 0.17033 -13.811 < 2e-16 ***
especePic vert -0.84328 0.14944 -5.643 1.67e-08 ***
especePie bavarde 0.91988 0.14511 6.339 2.31e-10 ***
especePigeon ramier 2.05308 0.15623 13.141 < 2e-16 ***
especePinson des arbres -1.25101 0.15284 -8.185 2.72e-16 ***
especePipit farlouse -1.20040 0.17727 -6.772 1.27e-11 ***
especePouillot véloce 0.08444 0.15808 0.534 0.59323
especeRoitelet à triple bandeau -0.36924 0.16001 -2.308 0.02102 *
especeRoitelet huppé -0.44197 0.17825 -2.480 0.01316 *
especeTroglodyte mignon 0.48798 0.15397 3.169 0.00153 **
especeVerdier d'Europe -2.16048 0.18060 -11.963 < 2e-16 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Threshold coefficients:
Estimate Std. Error z value
Déclin|Stable -1.63571 0.07555 -21.65
Stable|Augmentation 1.60172 0.07514 21.32
trait
df AIC
m_n1_trait_allint 24 12041.85
m_n1_trait_allint_1 24 12041.85
m_n1_trait_allint_2 24 12041.85
m_n1_trait_all 15 12449.00
m_n1_trait_3int 8 12626.41
m_n1_trait_3 6 12663.54
m_n1_trait_4int 10 12907.52
m_n1_trait_6int 8 13007.70
m_n1_trait_8int 8 13025.04
m_n1_trait_2 6 13155.55
m_n1_trait_5int 9 13187.16
m_n1_trait_7int 8 13207.88
m_n1_trait_8 6 13235.30
m_n1_trait_7 6 13279.68
m_n1_trait_4 7 13290.92
m_n1_trait_6 6 13329.04
m_n1_trait_5 7 13387.73
m_n1_trait_1 6 13436.64
m_n1_base 5 13464.97
Cumulative Link Mixed Model fitted with the Laplace approximation
formula:
avis_niv1 ~ tendance_shoc_niv1 + abondance_log + regime + popularite +
detectabilite + migrateur + gregaire + fauneFrance + pourcentage_stoc +
tendance_shoc_niv1:(abondance_log + popularite + detectabilite +
regime + migrateur + gregaire + fauneFrance + pourcentage_stoc) +
(1 | email)
data: dt1
link threshold nobs logLik AIC niter max.grad cond.H
logit flexible 7120 -5996.93 12041.85 4922(14871) 8.11e-03 5.1e+06
Random effects:
Groups Name Variance Std.Dev.
email (Intercept) 0.4857 0.6969
Number of groups: email 397
Coefficients:
Estimate Std. Error z value Pr(>|z|)
tendance_shoc_niv1.L -33.83330 4.39527 -7.698 1.39e-14
tendance_shoc_niv1.Q 7.94314 2.15689 3.683 0.000231
abondance_log -1.40453 0.22996 -6.108 1.01e-09
regimeInsectivore 3.01276 0.32411 9.295 < 2e-16
regimeOmnivore -0.16919 0.31217 -0.542 0.587836
populariteOui 4.73386 0.41746 11.340 < 2e-16
detectabiliteForte 2.26927 0.26397 8.597 < 2e-16
migrateurOui 3.13955 0.52430 5.988 2.12e-09
migrateurPartiel 6.69421 0.55947 11.965 < 2e-16
gregaireOui -4.31148 0.38067 -11.326 < 2e-16
fauneFrance -1.51377 0.21863 -6.924 4.40e-12
pourcentage_stoc 3.15279 0.74144 4.252 2.12e-05
tendance_shoc_niv1.L:abondance_log 3.74992 0.37986 9.872 < 2e-16
tendance_shoc_niv1.Q:abondance_log -1.32127 0.18979 -6.962 3.36e-12
tendance_shoc_niv1.L:populariteOui -5.94210 0.64051 -9.277 < 2e-16
tendance_shoc_niv1.Q:populariteOui 10.09106 1.07339 9.401 < 2e-16
tendance_shoc_niv1.L:detectabiliteForte -4.77114 1.00698 -4.738 2.16e-06
tendance_shoc_niv1.Q:detectabiliteForte 0.01013 0.38481 0.026 0.978990
tendance_shoc_niv1.L:regimeInsectivore -10.68331 0.84097 -12.704 < 2e-16
tendance_shoc_niv1.L:regimeOmnivore -4.34448 0.71153 -6.106 1.02e-09
tendance_shoc_niv1.L:migrateurPartiel -8.97309 0.73699 -12.175 < 2e-16
tendance_shoc_niv1.L ***
tendance_shoc_niv1.Q ***
abondance_log ***
regimeInsectivore ***
regimeOmnivore
populariteOui ***
detectabiliteForte ***
migrateurOui ***
migrateurPartiel ***
gregaireOui ***
fauneFrance ***
pourcentage_stoc ***
tendance_shoc_niv1.L:abondance_log ***
tendance_shoc_niv1.Q:abondance_log ***
tendance_shoc_niv1.L:populariteOui ***
tendance_shoc_niv1.Q:populariteOui ***
tendance_shoc_niv1.L:detectabiliteForte ***
tendance_shoc_niv1.Q:detectabiliteForte
tendance_shoc_niv1.L:regimeInsectivore ***
tendance_shoc_niv1.L:regimeOmnivore ***
tendance_shoc_niv1.L:migrateurPartiel ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Threshold coefficients:
Estimate Std. Error z value
Déclin|Stable -16.007 3.367 -4.755
Stable|Augmentation -12.770 3.363 -3.797
participant
df AIC
m_n1_participant_allint 14 13429.54
m_n1_participant_1int 8 13438.72
m_n1_participant_1 6 13450.92
m_n1_participant_2int 8 13451.01
m_n1_participant_all 8 13454.38
m_n1_participant_3int 8 13461.72
m_n1_base 5 13464.97
m_n1_participant_3 6 13465.59
m_n1_participant_2 6 13466.75
Cumulative Link Mixed Model fitted with the Laplace approximation
formula: avis_niv1 ~ tendance_shoc_niv1 + sexe + anciennete + expert +
tendance_shoc_niv1:(sexe + anciennete + expert) + (1 | email)
data: dt1
link threshold nobs logLik AIC niter max.grad cond.H
logit flexible 7120 -6700.77 13429.54 2444(7353) 3.28e-03 1.4e+03
Random effects:
Groups Name Variance Std.Dev.
email (Intercept) 0.3077 0.5547
Number of groups: email 397
Coefficients:
Estimate Std. Error z value Pr(>|z|)
tendance_shoc_niv1.L 0.6883376 0.0951473 7.234 4.67e-13 ***
tendance_shoc_niv1.Q -0.1260898 0.1076122 -1.172 0.241316
sexeH 0.3364358 0.0860621 3.909 9.26e-05 ***
anciennete 0.0001126 0.0060716 0.019 0.985205
expertOui 0.0404247 0.0848088 0.477 0.633606
tendance_shoc_niv1.L:sexeH 0.3406652 0.0906436 3.758 0.000171 ***
tendance_shoc_niv1.Q:sexeH -0.1036382 0.1040135 -0.996 0.319060
tendance_shoc_niv1.L:anciennete 0.0236736 0.0062633 3.780 0.000157 ***
tendance_shoc_niv1.Q:anciennete -0.0018711 0.0072067 -0.260 0.795142
tendance_shoc_niv1.L:expertOui 0.0549318 0.0886425 0.620 0.535455
tendance_shoc_niv1.Q:expertOui 0.1231838 0.1017438 1.211 0.226001
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Threshold coefficients:
Estimate Std. Error z value
Déclin|Stable -1.00159 0.09024 -11.10
Stable|Augmentation 1.77620 0.09232 19.24
plot (latente)

plot (response)
2 plot (response)
3 plot (response)
espece + participant
df AIC
m_n1_trait_participant_allint_3 31 11999.66
m_n1_trait_participant_allint_4 31 12010.71
m_n1_trait_participant_allint_2 29 12016.33
m_n1_trait_participant_allint_1 27 12028.18
m_n1_trait_allint 24 12041.85
m_n1_trait_participant_allint_6 28 12266.92
m_n1_trait_participant_allint_9 26 12360.86
m_n1_trait_participant_allint_8 26 12387.59
m_n1_trait_participant_allint_5 24 12406.78
m_n1_trait_participant_allint_7 24 12406.78
m_n1_trait_all 15 12449.00
m_n1_trait_3int 8 12626.41
m_n1_trait_3 6 12663.54
m_n1_trait_4int 10 12907.52
m_n1_trait_6int 8 13007.70
m_n1_trait_2int 8 13124.40
m_n1_trait_2 6 13155.55
m_n1_trait_5int 9 13187.16
m_n1_trait_4 7 13290.92
m_n1_trait_6 6 13329.04
m_n1_trait_1int 8 13348.01
m_n1_trait_5 7 13387.73
m_n1_participant_allint 14 13429.54
m_n1_trait_1 6 13436.64
m_n1_participant_1int 8 13438.72
m_n1_participant_1 6 13450.92
m_n1_participant_2int 8 13451.01
m_n1_participant_all 8 13454.38
m_n1_participant_3int 8 13461.72
m_n1_base 5 13464.97
m_n1_participant_3 6 13465.59
m_n1_participant_2 6 13466.75
Cumulative Link Mixed Model fitted with the Laplace approximation
formula:
avis_niv1 ~ tendance_shoc_niv1 + abondance_log + regime + popularite +
detectabilite + migrateur + gregaire + fauneFrance + pourcentage_stoc +
sexe + anciennete + expert + tendance_shoc_niv1:(abondance_log +
popularite + detectabilite + regime + migrateur + gregaire +
sexe + anciennete) + (1 | email)
data: dt1
link threshold nobs logLik AIC niter max.grad cond.H
logit flexible 7120 -5968.83 11999.66 7900(23704) 3.26e+00 7.0e+06
Random effects:
Groups Name Variance Std.Dev.
email (Intercept) 0.4518 0.6722
Number of groups: email 397
Coefficients:
Estimate Std. Error z value Pr(>|z|)
tendance_shoc_niv1.L -32.585556 4.612988 -7.064 1.62e-12
tendance_shoc_niv1.Q 8.214112 2.192154 3.747 0.000179
abondance_log -1.298261 0.239213 -5.427 5.72e-08
regimeInsectivore 3.005832 0.329281 9.128 < 2e-16
regimeOmnivore -0.049018 0.321235 -0.153 0.878720
populariteOui 4.690181 0.429352 10.924 < 2e-16
detectabiliteForte 2.305668 0.264569 8.715 < 2e-16
migrateurOui 3.258662 0.528465 6.166 6.99e-10
migrateurPartiel 6.508037 0.584195 11.140 < 2e-16
gregaireOui -4.200506 0.396084 -10.605 < 2e-16
fauneFrance -1.571077 0.221197 -7.103 1.22e-12
pourcentage_stoc 2.895107 0.766215 3.778 0.000158
sexeH 0.399897 0.098626 4.055 5.02e-05
anciennete 0.001510 0.006964 0.217 0.828382
expertOui 0.086680 0.096747 0.896 0.370282
tendance_shoc_niv1.L:abondance_log 3.603401 0.400169 9.005 < 2e-16
tendance_shoc_niv1.Q:abondance_log -1.313038 0.193832 -6.774 1.25e-11
tendance_shoc_niv1.L:populariteOui -5.826193 0.657463 -8.862 < 2e-16
tendance_shoc_niv1.Q:populariteOui 9.667994 1.126479 8.582 < 2e-16
tendance_shoc_niv1.L:detectabiliteForte -4.368360 1.044393 -4.183 2.88e-05
tendance_shoc_niv1.Q:detectabiliteForte 0.087217 0.387031 0.225 0.821708
tendance_shoc_niv1.L:regimeInsectivore -10.461893 0.875271 -11.953 < 2e-16
tendance_shoc_niv1.L:regimeOmnivore -4.479939 0.721440 -6.210 5.31e-10
tendance_shoc_niv1.L:migrateurPartiel -8.758598 0.771341 -11.355 < 2e-16
tendance_shoc_niv1.L:sexeH 0.368286 0.095573 3.853 0.000116
tendance_shoc_niv1.Q:sexeH -0.100292 0.108433 -0.925 0.355011
tendance_shoc_niv1.L:anciennete 0.027867 0.006182 4.508 6.55e-06
tendance_shoc_niv1.Q:anciennete 0.002904 0.007023 0.413 0.679285
tendance_shoc_niv1.L ***
tendance_shoc_niv1.Q ***
abondance_log ***
regimeInsectivore ***
regimeOmnivore
populariteOui ***
detectabiliteForte ***
migrateurOui ***
migrateurPartiel ***
gregaireOui ***
fauneFrance ***
pourcentage_stoc ***
sexeH ***
anciennete
expertOui
tendance_shoc_niv1.L:abondance_log ***
tendance_shoc_niv1.Q:abondance_log ***
tendance_shoc_niv1.L:populariteOui ***
tendance_shoc_niv1.Q:populariteOui ***
tendance_shoc_niv1.L:detectabiliteForte ***
tendance_shoc_niv1.Q:detectabiliteForte
tendance_shoc_niv1.L:regimeInsectivore ***
tendance_shoc_niv1.L:regimeOmnivore ***
tendance_shoc_niv1.L:migrateurPartiel ***
tendance_shoc_niv1.L:sexeH ***
tendance_shoc_niv1.Q:sexeH
tendance_shoc_niv1.L:anciennete ***
tendance_shoc_niv1.Q:anciennete
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Threshold coefficients:
Estimate Std. Error z value
Déclin|Stable -14.084 3.494 -4.031
Stable|Augmentation -10.835 3.490 -3.104
Interprétation
1️⃣ Ce que modélise exactement ce CLMM
Variable réponse : avis_niv1 → variable ordinale à 3 niveaux (Déclin < Stable < Augmentation).
Modèle : Cumulative Link Mixed Model → on modélise la probabilité cumulée d’être dans une catégorie plus élevée.
Effet aléatoire : (1 | email) → chaque répondant a une propension personnelle à répondre plus pessimiste/optimiste → variance = 0.45, non négligeable → bon choix de modèle mixte.
2️⃣ Comment lire un coefficient dans ce type de modèle
👉 Signe du coefficient
positif → augmente la probabilité d’un avis plus favorable (vers Augmentation)
négatif → pousse vers un avis plus défavorable (Déclin)
👉 Attention
Les coefficients sont sur l’échelle du log-odds cumulatif
Leur valeur absolue n’est pas directement intuitive
Les interactions sont centrales ici → les effets simples ne s’interprètent jamais seuls
3️⃣ Effet clé : tendance_shoc_niv1
Tu as codé cette variable comme ordinale polynomiale :
.L = tendance linéaire
.Q = tendance quadratique
Effets principaux tendance_shoc_niv1.L = -32.6 tendance_shoc_niv1.Q = +8.2
👉 Interprétation
Plus la tendance SHOC est perçue comme négative, plus l’avis est très fortement tiré vers le déclin
L’effet n’est pas linéaire : il y a une courbure importante (Q significatif)
⚠️ MAIS : ces effets sont massivement modulés par les interactions → ils ne valent que pour la modalité de référence de toutes les autres variables.
4️⃣ Effets principaux des espèces (hors interactions)
À lire comme des biais systématiques indépendants de SHOC :
Effets allant vers Augmentation
regimeInsectivore +++
populariteOui +++
detectabiliteForte +++
migrateurOui / Partiel +++
pourcentage_stoc +
sexeH +
👉 Ces espèces (ou répondants) sont structurellement jugés plus positivement
Effets allant vers Déclin
abondance_log —
gregaireOui —
fauneFrance —
👉 Plus une espèce est abondante, grégaire ou bien connue, plus l’avis est sévère → biais classique de baseline shifting / exigence accrue
Effets non significatifs
anciennete
expert
👉 Pas de différence nette d’avis entre experts et non-experts, ni selon l’ancienneté en moyenne
5️⃣ Le cœur du modèle : les interactions avec tendance_shoc_niv1
C’est là que ton modèle devient intéressant scientifiquement.
Exemple clé : abondance tendance.L : abondance_log +++ tendance.Q : abondance_log —
👉 Lecture
Quand la tendance SHOC devient négative :
les espèces abondantes sont encore plus pénalisées
effet non linéaire → les répondants semblent dire :
“Si même une espèce abondante décline, alors c’est très grave.”
Popularité tendance.L : populariteOui — tendance.Q : populariteOui +++
👉 Les espèces populaires :
bénéficient d’un effet de seuil
mais subissent une chute brutale quand SHOC est négatif
Régime alimentaire tendance.L : Insectivore — tendance.L : Omnivore —
👉 Les insectivores (et omnivores) :
sont beaucoup plus sensibles à une tendance SHOC négative
signal perçu comme écologiquement alarmant
Migration tendance.L : migrateurPartiel —
👉 Les migrateurs partiels :
subissent une surpénalisation dès que SHOC se dégrade
Sexe et ancienneté tendance.L : sexeH + tendance.L : anciennete +
👉 Les hommes et les répondants expérimentés :
réagissent plus fortement à la dégradation SHOC
mais pas de différence moyenne, seulement dans la réaction au signal
6️⃣ Seuils (thresholds) Déclin | Stable = -14.1 Stable | Augmentation = -10.8
👉 Ces valeurs indiquent que :
il faut une très forte combinaison d’effets positifs pour atteindre Augmentation
le modèle est asymétrique, avec un biais structurel vers Déclin
👉 Cohérent avec :
une variable réponse déséquilibrée
ou une perception globalement pessimiste
7️⃣ Message scientifique synthétique (si tu dois l’écrire)
Les avis ne dépendent pas seulement de la tendance observée (SHOC), mais de la cohérence perçue entre cette tendance et les caractéristiques écologiques de l’espèce. Une dégradation est jugée d’autant plus sévèrement qu’elle concerne une espèce abondante, populaire, insectivore ou migratrice, suggérant un raisonnement fondé sur la plausibilité écologique et la gravité implicite du signal.
niv 2
données
aléatoire
df AIC
m_n2.1 7 18975.04
m_n2.7 8 18976.87
m_n2.6 8 18977.04
m_n2.5 8 18977.04
m_n2.4 7 19157.90
m_n2.2 7 19197.78
m_n2.3 7 19201.27
Cumulative Link Mixed Model fitted with the Laplace approximation
formula: avis_niv2 ~ 1 + (1 | email)
data: dt2
link threshold nobs logLik AIC niter max.grad cond.H
logit flexible 6881 -9480.52 18975.04 376(1557) 4.89e-03 2.8e+01
Random effects:
Groups Name Variance Std.Dev.
email (Intercept) 0.3393 0.5825
Number of groups: email 397
No Coefficients
Threshold coefficients:
Estimate Std. Error z value
Déclin fort|Déclin modéré à fort -3.71827 0.08099 -45.91
Déclin modéré à fort|Déclin modéré -2.35538 0.05194 -45.35
Déclin modéré|Stable -1.20747 0.04182 -28.87
Stable|Augmentation modéré 1.43182 0.04298 33.31
Augmentation modéré|Augmentation modéré à fort 2.75992 0.05775 47.79
Augmentation modéré à fort|Augmentation fort 4.06019 0.09279 43.76
base
espece
Cumulative Link Mixed Model fitted with the Laplace approximation
formula: avis_niv2 ~ tendance_shoc_niv2 * espece + (1 | email) + (1 |
departement)
data: dt2
link threshold nobs logLik AIC niter max.grad cond.H
logit flexible 6881 -8246.69 16551.38 5869(29226) 2.63e-03 1.2e+02
Random effects:
Groups Name Variance Std.Dev.
email (Intercept) 0.53687 0.7327
departement (Intercept) 0.01533 0.1238
Number of groups: email 397, departement 89
Coefficients:
Estimate Std. Error z value Pr(>|z|)
tendance_shoc_niv2.L 2.59038 0.11813 21.929 < 2e-16 ***
tendance_shoc_niv2.Q -2.02970 0.10978 -18.488 < 2e-16 ***
tendance_shoc_niv2.C -0.13073 0.11483 -1.139 0.25490
tendance_shoc_niv2^4 0.09672 0.12128 0.797 0.42519
tendance_shoc_niv2^5 0.25080 0.11509 2.179 0.02932 *
tendance_shoc_niv2^6 -0.82591 0.10985 -7.519 5.53e-14 ***
especeBouscarle de Cetti 1.75330 0.17436 10.056 < 2e-16 ***
especeBruant zizi -1.20415 0.18184 -6.622 3.54e-11 ***
especeChoucas des tours 0.96913 0.16885 5.739 9.50e-09 ***
especeGeai des chênes 1.43647 0.15460 9.291 < 2e-16 ***
especeGrive draine -0.49617 0.15844 -3.132 0.00174 **
especeMerle noir 1.05876 0.15432 6.861 6.84e-12 ***
especeMésange bleue -0.10694 0.15030 -0.712 0.47675
especeMésange nonnette -2.37066 0.16731 -14.169 < 2e-16 ***
especePic vert -0.86119 0.15117 -5.697 1.22e-08 ***
especePie bavarde 0.96766 0.14616 6.621 3.57e-11 ***
especePigeon ramier 2.36308 0.16495 14.326 < 2e-16 ***
especePinson des arbres -1.29323 0.15538 -8.323 < 2e-16 ***
especePipit farlouse -1.30592 0.16358 -7.983 1.42e-15 ***
especeRoitelet à triple bandeau -0.30386 0.16232 -1.872 0.06121 .
especeTroglodyte mignon 0.49905 0.15280 3.266 0.00109 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Threshold coefficients:
Estimate Std. Error z value
Déclin fort|Déclin modéré à fort -4.09780 0.09875 -41.49
Déclin modéré à fort|Déclin modéré -2.50228 0.07273 -34.41
Déclin modéré|Stable -1.02940 0.06377 -16.14
Stable|Augmentation modéré 2.35574 0.06854 34.37
Augmentation modéré|Augmentation modéré à fort 3.94700 0.08301 47.55
Augmentation modéré à fort|Augmentation fort 5.42180 0.11439 47.40
trait
df AIC
m_n2_trait_allint_7 29 16551.78
m_n2_trait_allint_2 27 16742.23
m_n2_trait_allint_3 26 16745.00
m_n2_trait_all 24 16817.16
m_n2_trait_4int 19 16961.88
m_n2_trait_3int 17 17006.53
m_n2_trait_3 15 17022.57
m_n2_trait_4 15 17037.68
m_n2_trait_7int 18 17163.13
m_n2_trait_6int 19 17220.58
m_n2_trait_8int 18 17321.68
m_n2_trait_9int 18 17321.68
m_n2_trait_5 16 17524.98
m_n2_trait_5int 19 17527.17
m_n2_trait_6 16 17543.66
m_n2_trait_2 15 17666.12
Cumulative Link Mixed Model fitted with the Laplace approximation
formula:
avis_niv2 ~ tendance_shoc_niv2 + abondance_log + regime + popularite +
detectabilite + migrateur + gregaire + fauneFrance + tendance_shoc_niv2:(popularite +
fauneFrance) + (1 | email) + (1 | departement)
data: dt2
link threshold nobs logLik AIC niter max.grad cond.H
logit flexible 6881 -8246.89 16551.78 7668(38554) 1.62e+00 1.6e+07
Random effects:
Groups Name Variance Std.Dev.
email (Intercept) 0.53539 0.7317
departement (Intercept) 0.01483 0.1218
Number of groups: email 397, departement 89
Coefficients:
Estimate Std. Error z value Pr(>|z|)
tendance_shoc_niv2.L 18.6679 4.6245 4.037 5.42e-05 ***
tendance_shoc_niv2.Q -5.2839 3.6430 -1.450 0.14694
tendance_shoc_niv2.C 5.0884 1.9554 2.602 0.00926 **
tendance_shoc_niv2^4 1.1865 0.4940 2.402 0.01632 *
tendance_shoc_niv2^5 1.1605 0.8884 1.306 0.19145
tendance_shoc_niv2^6 -0.3956 0.2684 -1.474 0.14053
abondance_log -2.0227 0.3434 -5.890 3.87e-09 ***
regimeInsectivore 2.3546 0.3210 7.336 2.20e-13 ***
regimeOmnivore -0.2957 0.2395 -1.235 0.21700
populariteOui 0.2289 1.0288 0.222 0.82395
detectabiliteForte 3.1847 0.3056 10.421 < 2e-16 ***
migrateurOui -4.6840 0.6821 -6.867 6.54e-12 ***
migrateurPartiel 0.9339 0.2371 3.939 8.19e-05 ***
gregaireOui -1.3387 0.3114 -4.299 1.71e-05 ***
fauneFrance 1.0872 1.9593 0.555 0.57896
tendance_shoc_niv2.L:populariteOui -9.9793 3.1150 -3.204 0.00136 **
tendance_shoc_niv2.Q:populariteOui 6.6385 3.9824 1.667 0.09553 .
tendance_shoc_niv2.L:fauneFrance 25.4603 5.9679 4.266 1.99e-05 ***
tendance_shoc_niv2.Q:fauneFrance 1.7427 6.1765 0.282 0.77782
tendance_shoc_niv2.C:fauneFrance 13.8755 2.8637 4.845 1.26e-06 ***
tendance_shoc_niv2^4:fauneFrance 2.4535 1.6742 1.465 0.14280
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Threshold coefficients:
Estimate Std. Error z value
Déclin fort|Déclin modéré à fort -30.017 4.678 -6.416
Déclin modéré à fort|Déclin modéré -28.423 4.677 -6.077
Déclin modéré|Stable -26.952 4.676 -5.764
Stable|Augmentation modéré -23.572 4.671 -5.046
Augmentation modéré|Augmentation modéré à fort -21.982 4.671 -4.706
Augmentation modéré à fort|Augmentation fort -20.510 4.672 -4.390
participant
df AIC
m_n2_participant_allint 35 17639.89
m_n2_participant_2int 21 17668.12
m_n2_participant_1int 21 17669.90
m_n2_participant_1 15 17687.81
m_n2_participant_3int 21 17688.90
m_n2_participant_all 17 17691.66
m_n2_base 14 17704.51
m_n2_participant_3 15 17705.85
m_n2_participant_2 15 17706.51
Cumulative Link Mixed Model fitted with the Laplace approximation
formula: avis_niv2 ~ tendance_shoc_niv2 + sexe + anciennete + expert +
tendance_shoc_niv2:(sexe + anciennete + expert) + (1 | email) +
(1 | departement)
data: dt2
link threshold nobs logLik AIC niter max.grad cond.H
logit flexible 6881 -8784.95 17639.89 9838(40019) 7.20e+00 1.1e+04
Random effects:
Groups Name Variance Std.Dev.
email (Intercept) 0.400634 0.6330
departement (Intercept) 0.005013 0.0708
Number of groups: email 397, departement 89
Coefficients:
Estimate Std. Error z value Pr(>|z|)
tendance_shoc_niv2.L 1.7065315 0.2358617 7.235 4.64e-13 ***
tendance_shoc_niv2.Q -1.4157355 0.1936547 -7.311 2.66e-13 ***
tendance_shoc_niv2.C -0.2342217 0.2172952 -1.078 0.2811
tendance_shoc_niv2^4 -0.0984195 0.2349298 -0.419 0.6753
tendance_shoc_niv2^5 0.0374411 0.1973759 0.190 0.8495
tendance_shoc_niv2^6 0.2132680 0.1412547 1.510 0.1311
sexeH 0.4180348 0.1051412 3.976 7.01e-05 ***
anciennete -0.0024701 0.0071525 -0.345 0.7298
expertOui 0.0353984 0.1027948 0.344 0.7306
tendance_shoc_niv2.L:sexeH 0.5537311 0.2193513 2.524 0.0116 *
tendance_shoc_niv2.Q:sexeH 0.0350449 0.1822404 0.192 0.8475
tendance_shoc_niv2.C:sexeH -0.1551090 0.2024632 -0.766 0.4436
tendance_shoc_niv2^4:sexeH -0.3286177 0.2172758 -1.512 0.1304
tendance_shoc_niv2^5:sexeH -0.0923026 0.1842655 -0.501 0.6164
tendance_shoc_niv2^6:sexeH -0.3189563 0.1352824 -2.358 0.0184 *
tendance_shoc_niv2.L:anciennete 0.0656225 0.0145023 4.525 6.04e-06 ***
tendance_shoc_niv2.Q:anciennete 0.0158704 0.0125117 1.268 0.2046
tendance_shoc_niv2.C:anciennete 0.0047300 0.0131641 0.359 0.7194
tendance_shoc_niv2^4:anciennete -0.0001457 0.0136071 -0.011 0.9915
tendance_shoc_niv2^5:anciennete -0.0090184 0.0117261 -0.769 0.4418
tendance_shoc_niv2^6:anciennete -0.0135529 0.0089772 -1.510 0.1311
tendance_shoc_niv2.L:expertOui 0.5024869 0.2140408 2.348 0.0189 *
tendance_shoc_niv2.Q:expertOui -0.0900615 0.1812601 -0.497 0.6193
tendance_shoc_niv2.C:expertOui 0.3931426 0.1956267 2.010 0.0445 *
tendance_shoc_niv2^4:expertOui -0.1582256 0.2058523 -0.769 0.4421
tendance_shoc_niv2^5:expertOui 0.1553770 0.1749548 0.888 0.3745
tendance_shoc_niv2^6:expertOui 0.0229809 0.1303826 0.176 0.8601
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Threshold coefficients:
Estimate Std. Error z value
Déclin fort|Déclin modéré à fort -3.6070 0.1339 -26.937
Déclin modéré à fort|Déclin modéré -2.0669 0.1163 -17.777
Déclin modéré|Stable -0.7265 0.1117 -6.503
Stable|Augmentation modéré 2.2839 0.1147 19.916
Augmentation modéré|Augmentation modéré à fort 3.7152 0.1224 30.364
Augmentation modéré à fort|Augmentation fort 5.0618 0.1429 35.429
plot

espece + participant
df AIC
m_n2_trait_participant_allint_11 42 16503.30
m_n2_trait_participant_allint_10 48 16503.35
m_n2_trait_participant_allint_12 36 16545.77
m_n2_trait_allint_7 29 16551.78
m_n2_trait_participant_allint_6 29 16643.38
m_n2_trait_participant_allint_9 46 16738.75
m_n2_trait_allint_2 27 16742.23
m_n2_trait_allint_3 26 16745.00
m_n2_trait_participant_allint_5 28 16789.74
m_n2_trait_all 24 16817.16
m_n2_trait_participant_all 26 16817.53
m_n2_trait_4int 19 16961.88
m_n2_trait_3int 17 17006.53
m_n2_trait_3 15 17022.57
m_n2_trait_4 15 17037.68
m_n2_trait_6int 19 17220.58
m_n2_trait_5 16 17524.98
m_n2_trait_5int 19 17527.17
m_n2_trait_6 16 17543.66
m_n2_trait_2 15 17666.12
m_n2_participant_2int 21 17668.12
m_n2_participant_1int 21 17669.90
m_n2_participant_1 15 17687.81
m_n2_participant_3int 21 17688.90
m_n2_participant_all 17 17691.66
m_n2_base 14 17704.51
m_n2_participant_3 15 17705.85
m_n2_participant_2 15 17706.51
Cumulative Link Mixed Model fitted with the Laplace approximation
formula:
avis_niv2 ~ tendance_shoc_niv2 + abondance_log + regime + popularite +
detectabilite + migrateur + gregaire + fauneFrance + sexe +
anciennete + expert + tendance_shoc_niv2:(sexe + anciennete +
fauneFrance) + (1 | email) + (1 | departement)
data: dt2
link threshold nobs logLik AIC niter max.grad cond.H
logit flexible 6881 -8209.65 16503.30 11492(56810) 1.03e+00 6.9e+06
Random effects:
Groups Name Variance Std.Dev.
email (Intercept) 0.49808 0.7057
departement (Intercept) 0.01047 0.1023
Number of groups: email 397, departement 89
Coefficients:
Estimate Std. Error z value Pr(>|z|)
tendance_shoc_niv2.L 1.496e+01 1.733e+00 8.632 < 2e-16 ***
tendance_shoc_niv2.Q -8.107e+00 1.651e+00 -4.909 9.15e-07 ***
tendance_shoc_niv2.C 3.218e+00 4.644e-01 6.929 4.24e-12 ***
tendance_shoc_niv2^4 1.059e+00 3.692e-01 2.869 0.004123 **
tendance_shoc_niv2^5 -1.583e+00 3.252e-01 -4.867 1.13e-06 ***
tendance_shoc_niv2^6 8.807e-01 2.189e-01 4.022 5.76e-05 ***
abondance_log -8.287e-01 1.239e-01 -6.688 2.26e-11 ***
regimeInsectivore 1.461e+00 2.353e-01 6.210 5.30e-10 ***
regimeOmnivore 3.158e-01 9.829e-02 3.213 0.001315 **
populariteOui -6.986e-01 2.089e-01 -3.344 0.000825 ***
detectabiliteForte 1.928e+00 1.949e-01 9.894 < 2e-16 ***
migrateurOui -2.633e+00 4.694e-01 -5.609 2.04e-08 ***
migrateurPartiel -1.358e-01 1.058e-01 -1.284 0.199161
gregaireOui -1.026e+00 9.781e-02 -10.490 < 2e-16 ***
fauneFrance -6.268e+00 1.328e+00 -4.720 2.35e-06 ***
sexeH 4.494e-01 1.126e-01 3.991 6.57e-05 ***
anciennete -7.468e-04 7.622e-03 -0.098 0.921947
expertOui 6.513e-02 1.006e-01 0.647 0.517494
tendance_shoc_niv2.L:sexeH 6.490e-01 2.229e-01 2.911 0.003600 **
tendance_shoc_niv2.Q:sexeH -3.843e-02 1.851e-01 -0.208 0.835543
tendance_shoc_niv2.C:sexeH -1.533e-01 2.060e-01 -0.744 0.456823
tendance_shoc_niv2^4:sexeH -3.300e-01 2.212e-01 -1.492 0.135829
tendance_shoc_niv2^5:sexeH -1.337e-01 1.877e-01 -0.712 0.476495
tendance_shoc_niv2^6:sexeH -3.118e-01 1.377e-01 -2.264 0.023577 *
tendance_shoc_niv2.L:anciennete 8.436e-02 1.378e-02 6.123 9.17e-10 ***
tendance_shoc_niv2.Q:anciennete 1.163e-02 1.188e-02 0.979 0.327597
tendance_shoc_niv2.C:anciennete 1.330e-02 1.252e-02 1.062 0.288190
tendance_shoc_niv2^4:anciennete -7.613e-03 1.294e-02 -0.588 0.556372
tendance_shoc_niv2^5:anciennete -1.014e-02 1.117e-02 -0.908 0.364088
tendance_shoc_niv2^6:anciennete -1.228e-02 8.567e-03 -1.433 0.151722
tendance_shoc_niv2.L:fauneFrance 3.512e+01 4.041e+00 8.692 < 2e-16 ***
tendance_shoc_niv2.Q:fauneFrance -2.120e+01 4.161e+00 -5.094 3.51e-07 ***
tendance_shoc_niv2.C:fauneFrance 1.788e+01 1.896e+00 9.428 < 2e-16 ***
tendance_shoc_niv2^4:fauneFrance -4.357e+00 1.038e+00 -4.197 2.70e-05 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Threshold coefficients:
Estimate Std. Error z value
Déclin fort|Déclin modéré à fort -11.814 2.046 -5.775
Déclin modéré à fort|Déclin modéré -10.212 2.044 -4.995
Déclin modéré|Stable -8.739 2.044 -4.276
Stable|Augmentation modéré -5.356 2.041 -2.624
Augmentation modéré|Augmentation modéré à fort -3.745 2.041 -1.835
Augmentation modéré à fort|Augmentation fort -2.251 2.042 -1.102
ACP
niv 1
set.seed(42)
# n <- 100 # Nombre d'observations
ACP_dt1 <- dataset %>%
filter(avis_niv1 %in% c("Déclin", "Stable", "Augmentation")) %>%
dplyr::select(avis_niv1,abondance_log,fauneFrance,anciennete,pourcentage_stoc,popularite,detectabilite,
regime,migrateur,gregaire,sexe,expert,bioregion) %>%
na.omit()
# 1. Préparation des données : encodage des variables catégorielles
# On utilise dummy coding pour les variables catégorielles (sauf "avis_niv1")
ACP_dt_dummy <- ACP_dt1 %>%
mutate(across(c(popularite, detectabilite, regime, migrateur, gregaire, sexe, expert, bioregion), as.factor)) %>%
fastDummies::dummy_cols(select_columns = c("popularite", "detectabilite", "regime", "migrateur", "gregaire", "sexe",
"expert", "bioregion")) %>%
dplyr::select(-"abondance_log",-"fauneFrance",-"anciennete",-"pourcentage_stoc",-"popularite",-"detectabilite",-"regime",
-"migrateur",-"gregaire",-"sexe",-"expert",-"bioregion")
# On sépare les variables continues et les variables catégorielles encodées
X <- ACP_dt_dummy %>%
dplyr::select(-avis_niv1) # On exclut "avis_niv1" pour l'ACP
# 2. ACP avec FactoMineR
# On utilise "avis_niv1" comme variable illustrative
res.pca <- FactoMineR::PCA(X, scale.unit = TRUE, ncp = 5, graph = FALSE)
# 3. Visualisation des résultats
# a) Projection des individus (colorés par "avis_niv1")
fviz_pca_ind(res.pca,
col.ind = as.factor(ACP_dt1$avis_niv1),
addEllipses = TRUE,
title = "Projection des individus (ACP)")Registered S3 methods overwritten by 'broom':
method from
nobs.fitdistr MuMIn
nobs.multinom MuMIn

fviz_pca_ind(res.pca, axes = c(1, 3), col.ind = as.factor(ACP_dt1$avis_niv1), addEllipses = TRUE)
fviz_pca_ind(res.pca, axes = c(2, 3), col.ind = as.factor(ACP_dt1$avis_niv1), addEllipses = TRUE)
# b) Cercle des corréla# b) Cercle des corréla# b) Cercle des corrélations (variables continues uniquement)
fviz_pca_var(res.pca,
col.var = "black",
title = "Cercle des corrélations (variables continues)")
# c) Variance expliquée
fviz_eig(res.pca,
addlines = TRUE,
ylim = c(0, 50),
title = "Variance expliquée par les composantes principales")
# d) Contribution des variables catégorielles (si besoin)
# Pour voir la contribution des modalités des variables catégorielles, on peut utiliser :
fviz_contrib(res.pca, choice = "var", axes = 1)
fviz_contrib(res.pca, choice = "var", axes = 2)
COnclusion
Les participants n’associent pas une perception de tendances de populations en fonction des traits des espèces ni en fonction de leur caractéristiques personnelles.
La perceptions des participants est relativement indépendantes des variables testées.
COLLECTIF
variables
avis_niv1 = “Déclin” ~ -1,“Augmentation” ~ 1, “Stable” ~ 0
round
ou
mean avis <= -0.2 ~ “Déclin”, mean avis >= 0.2 ~ “Augmentation”, -0.2 < mean avis > 0.2 ~ “Stable”
-1 0 1 <NA>
3269 5003 2033 0
Chi² : shoc vs round
Augmentation Déclin Stable
SHOC 9 8 5
AVIS 1 5 22
Warning in chisq.test(tab_chi2_round): L’approximation du Chi-2 est peut-être
incorrecte
Pearson's Chi-squared test
data: tab_chi2_round
X-squared = 17.325, df = 2, p-value = 0.0001729
Warning in chisq.test(tab_chi2_round): L’approximation du Chi-2 est peut-être
incorrecte
[1] "résidus"
Augmentation Déclin Stable
SHOC 3.276367 1.480904 -3.932849
AVIS -3.276367 -1.480904 3.932849
[1] "proportions"
Augmentation Déclin Stable
SHOC 0.40909091 0.3636364 0.2272727
AVIS 0.03571429 0.1785714 0.7857143
Chi² : shoc vs quantile 0.2
Augmentation Déclin Stable
SHOC 9 8 5
AVIS 6 11 11
Pearson's Chi-squared test
data: tab_chi2_quantile02
X-squared = 2.6417, df = 2, p-value = 0.2669
[1] "résidus"
Augmentation Déclin Stable
SHOC 1.492094 -0.2113049 -1.245935
AVIS -1.492094 0.2113049 1.245935
[1] "proportions"
Augmentation Déclin Stable
SHOC 0.4090909 0.3636364 0.2272727
AVIS 0.2142857 0.3928571 0.3928571
tendances par espece
mean_intel_dt <- mean_intel_dt %>%
left_join(tendance_shoc)Joining with `by = join_by(espece)`
dt_intel_quantile02 <- mean_intel_dt %>%
dplyr::select(espece, round_avis_niv1_intel_cat, quantile02_avis_niv1_intel_cat, tendance_shoc_niv1) %>%
rename(espece = espece,
`tendance moyenne arrondie quiz` = round_avis_niv1_intel_cat,
`tendance moyenne quantile (-0.2, 0.2) quiz` = quantile02_avis_niv1_intel_cat,
`tendance stat SHOC` = tendance_shoc_niv1)
dt_intel_quantile02 %>%
datatable(extensions = 'Buttons',
options = list(dom = 'Blfrtip',
buttons = c('copy', 'csv', 'excel', 'pdf', 'print'),
lengthMenu = list(c(10,25,50,-1),
c(10,25,50,"All"))))dt_intel_quantile02# A tibble: 28 × 4
espece tendance moyenne arr…¹ tendance moyenne qua…² `tendance stat SHOC`
<chr> <chr> <chr> <chr>
1 Accenteur… Stable Déclin Déclin
2 Alouette … Déclin Déclin Incertain
3 Bouscarle… Stable Augmentation Augmentation
4 Bruant zi… Stable Stable Augmentation
5 Chardonne… Stable Déclin Incertain
6 Choucas d… Stable Augmentation Augmentation
7 Geai des … Stable Stable Déclin
8 Grimperea… Stable Stable Augmentation
9 Grive dra… Stable Déclin Déclin
10 Grive lit… Déclin Déclin Incertain
# ℹ 18 more rows
# ℹ abbreviated names: ¹`tendance moyenne arrondie quiz`,
# ²`tendance moyenne quantile (-0.2, 0.2) quiz`
variabilité des réponses par espece
Pas de correlation entre le tendances moyennes estimées collectivement et la variation (écart-type) dans les réponses
plot(mean_intel_dt$mean_avis_niv1_intel, mean_intel_dt$sd_avis_niv1_intel)
summary(lm(mean_intel_dt$mean_avis_niv1_intel ~ mean_intel_dt$sd_avis_niv1_intel))
Call:
lm(formula = mean_intel_dt$mean_avis_niv1_intel ~ mean_intel_dt$sd_avis_niv1_intel)
Residuals:
Min 1Q Median 3Q Max
-0.5019 -0.3525 0.0251 0.3158 0.8052
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -0.8524 0.4731 -1.802 0.0832 .
mean_intel_dt$sd_avis_niv1_intel 1.1782 0.7886 1.494 0.1472
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 0.3857 on 26 degrees of freedom
Multiple R-squared: 0.07907, Adjusted R-squared: 0.04365
F-statistic: 2.232 on 1 and 26 DF, p-value: 0.1472